Template Library Catalogue

Research and architecture documents

• Shared Governance Kernel
• Shared Specialist Control Registry
• Asset Metadata Schema
• Migration Notes
• Scientific DNA
• Prompt Architecture Guide
• Traceability Matrix Pattern
• System 2 Prompting Guide
• Research and Standards Basis
• Curated Research Source Registry
• Human-AI Quality Standard

Research control docs

• RESEARCH-DNA.md
• PROMPT-PATTERN-MATRIX.md
• SCIENTIFIC-CONTROL-CHECKLIST.md

This catalogue contains 100 named assets: 40 prompts, 30 skills, and 30 contracts.

Every asset is scenario-specific, copy-ready, and written for evidence-first agent work, failure disclosure, verification, and reduced completion theatre.

Prompts

Core Control

• Master Agent Enforcement Prompt — Use at the start of a serious coding-agent session when the agent must obey repository rules, evidence discipline, and release-honesty constraints.

Implementation

• Evidence-First Implementation Prompt — Use when a feature, bug fix, refactor, or documentation change must be implemented with proof before any success claim.

Audit

• Anti-Fake Completion Audit Prompt — Use after an agent claims success and you need to test whether the claim is real, exaggerated, or unsupported.

Reasoning Control

• Assumption Exposure Prompt — Use before implementation or review when hidden assumptions could change architecture, UX, security, or release judgement.
• Uncertainty Calibration Prompt — Use when an answer sounds confident but evidence quality varies across code, UI, tests, or documentation.
• Contradiction Detection Prompt — Use when docs, tests, UI, code, README, or release language may disagree.

Risk

• Premortem Failure Analysis Prompt — Use before merge, release, or large refactor to imagine failure and prevent avoidable blind spots.

Scoping

• Scope Boundary Lock Prompt — Use when the task is at risk of scope creep, hidden dependencies, or accidental redesign.

Implementation

• Minimal Correct Change Prompt — Use when a requested fix must avoid broad rewrites, unrelated styling, or unnecessary dependency changes.

Repository Analysis

• Repository Reconnaissance Prompt — Use at the beginning of work on an unfamiliar repository, template pack, or docs/product system.

Implementation

• Feature Implementation Prompt — Use for new feature work that needs design, code, tests, docs, and release evidence kept aligned.

Bug Remediation

• Bug Root Cause Remediation Prompt — Use when a symptom is known but the actual cause, blast radius, and regression risk must be proven.

Review

• Regression Risk Review Prompt — Use after any implementation that might break adjacent behaviour, shared contracts, or existing user paths.

Code Quality

• Code Quality Deep Audit Prompt — Use for maintainability, structure, dead-code, duplication, and readability review.

Architecture

• Architecture Drift Audit Prompt — Use when a change may violate boundaries, patterns, ownership, or long-term design constraints.

Clean-up

• Dead Code Removal Prompt — Use when removing unused files, branches, exports, styles, docs, tests, or generated artefacts.

Refactor

• Refactor Safety Prompt — Use when improving structure without intentionally changing behaviour.

API

• API Contract Review Prompt — Use when request/response shapes, public functions, schemas, integrations, or API docs may change.

Data Flow

• Data Flow Trace Prompt — Use when data origin, transformation, persistence, display, export, or logging must be understood.

State

• State Management Audit Prompt — Use for frontend or application state that may become stale, inconsistent, duplicated, or race-prone.

UI/UX

• UI Structure Audit Prompt — Use when a UI must be inspected for hierarchy, grouping, spacing, layout, and real rendered structure.
• Visual Design Quality Gate Prompt — Use before accepting a UI that must look polished, readable, aligned, and professionally finished.

Design System

• Design System Consistency Prompt — Use when components, tokens, typography, icons, spacing, and interaction states must remain consistent.

Responsive UI

• Responsive Layout Audit Prompt — Use when layouts must survive mobile, tablet, desktop, long content, and constrained viewport states.

Interaction

• Interaction State Audit Prompt — Use for buttons, forms, menus, tables, cards, dialogs, navigation, and other interactive UI states.

Accessibility

• Keyboard Journey Audit Prompt — Use when interactive surfaces must be navigable without a mouse and must avoid keyboard traps.
• Accessibility Remediation Prompt — Use when a specific accessibility issue must be fixed without fake pass language or harmful ARIA patches.
• Screen Reader Risk Review Prompt — Use when dynamic UI, forms, icon controls, modals, route changes, or live regions may confuse assistive technology users.
• WCAG Mapping Review Prompt — Use when findings need a careful mapping to WCAG 2.2 success criteria and severity without overstating certainty.
• Icon-Only Control Review Prompt — Use when icon buttons, compact actions, menus, or toolbar controls may lack clear names or affordances.

Testing

• Playwright Test Design Prompt — Use when designing browser tests for real user behaviour, accessibility checks, and stable selectors.
• Test Quality Audit Prompt — Use when tests pass but may be shallow, flaky, over-mocked, or unrelated to real risk.

Release

• Release Evidence Prompt — Use before final handoff, merge approval, release note creation, or claim publication.

Documentation

• Documentation Truthfulness Prompt — Use when docs, guides, examples, or site copy may promise more than the implementation proves.
• README Upgrade Prompt — Use when a README needs to become a useful product entry point, not just a file list.

Site/Product

• GitHub Pages Template Hub Prompt — Use when turning repository files into a public template hub with copy actions, categories, examples, and low-friction navigation.

Security

• Security Trust Boundary Prompt — Use when a change touches actors, permissions, data, integration, exports, admin features, or sensitive operations.
• Secrets Handling Review Prompt — Use when environment variables, tokens, credentials, logs, CI, config, or integrations may expose secrets.

Observability

• Observability and Logging Prompt — Use when work needs useful logs, metrics, traces, alerts, or operational evidence without privacy leakage.

Release

• Final Merge Gate Prompt — Use as the final review prompt before merge or publication when all evidence must be consolidated.

Skills

Scoping

• Scoping Packet Skill — Use when converting a task request into a bounded, reviewable scoping packet.

Planning

• Implementation Planning Skill — Use when the agent needs a safe implementation plan before editing.

Repository Analysis

• Repository Mapping Skill — Use when the repository structure, authority files, and scripts must be understood before change.

Evidence

• Evidence Collection Skill — Use when gathering proof for implementation, review, documentation, or release claims.

Verification

• Verification Reporting Skill — Use when command results, manual checks, and skipped checks must be reported cleanly.

Honesty

• Failure Disclosure Skill — Use when the answer must disclose incomplete, unverified, risky, or weaker-than-requested work.

UI/UX

• UI Screenshot Review Skill — Use when screenshots or rendered UI must be audited for defects and state mismatches.

Accessibility

• Accessibility Inspection Skill — Use when inspecting semantic structure, names, keyboard behaviour, and user impact.
• WCAG Mapping Skill — Use when translating accessibility findings into WCAG 2.2 references and remediation evidence.
• Keyboard Testing Skill — Use when interactive flows must be tested by keyboard.

Testing

• Playwright Debugging Skill — Use when browser tests fail, flake, or need meaningful behaviour assertions.
• Regression Test Design Skill — Use when a bug fix or feature needs tests that would catch future breakage.

Review

• Code Review Skill — Use when reviewing a diff for correctness, maintainability, and evidence.

Refactor

• Refactor Planning Skill — Use when structural improvement is needed but behaviour must stay stable.

Clean-up

• Dead Code Detection Skill — Use when deciding whether files, exports, branches, styles, or docs are unused.

Architecture

• Architecture Review Skill — Use when reviewing boundaries, ownership, coupling, and long-term maintainability.

API

• API Contract Inspection Skill — Use when API shapes, schemas, callers, and docs must be reconciled.

Security

• Security Review Skill — Use when reviewing trust boundaries, permissions, secrets, data handling, and abuse paths.
• Secrets Review Skill — Use when tokens, env vars, logs, config, or CI may expose secrets.

Privacy

• Data Privacy Review Skill — Use when personal, tenant, customer, analytics, export, or audit data may be collected or exposed.

Documentation

• Documentation Audit Skill — Use when docs must be checked against implementation and repository reality.

Release

• Release Notes Review Skill — Use when release notes must represent verified behaviour and known limitations.

Documentation

• README Structuring Skill — Use when README needs a clearer user journey, catalogue, and verification instructions.

Template Product

• Template Catalogue Design Skill — Use when organising many templates into categories, starter packs, and discoverable metadata.

Site/Product

• GitHub Pages UX Skill — Use when the public site must serve templates clearly on GitHub Pages.

Prompt Writing

• Copy-Ready Prompt Writing Skill — Use when prompts must be immediately usable by agents without vague placeholders.

Agent Control

• Agent Behaviour Hardening Skill — Use when prompts or workflows need stronger controls against shortcuts and overclaiming.

Evaluation

• Prompt Evaluation Skill — Use when assessing whether prompts produce reliable, honest, and useful agent output.

Risk

• Risk Register Skill — Use when work requires a live list of risks, owners, mitigations, and verification gaps.

Delivery

• Final Delivery Packaging Skill — Use when packaging an upgraded repository, template system, or release bundle.

Contracts

Honesty

• Completion Honesty Contract — Use as the top-level acceptance rule against fake completion.

Evidence

• Evidence Contract — Use when every claim must be traceable to inspected files, behaviour, commands, or manual evidence.

Scope

• Scope Control Contract — Use to reject hidden scope expansion, accidental redesign, or unrelated changes.

Reasoning Control

• Assumption Contract — Use when assumptions must be named before they influence implementation or review.

Verification

• Verification Contract — Use when commands, manual checks, and evidence gaps define acceptance.

Testing

• Test Behaviour Contract — Use when tests must prove user or system behaviour rather than shallow implementation details.

Accessibility

• Accessibility Contract Plus — Use for UI work that must preserve semantic structure, keyboard access, and WCAG-aligned reasoning.
• Keyboard Interaction Contract — Use for menus, dialogs, forms, tables, navigation, and custom controls.
• Accessible Name Contract — Use for icon controls, form controls, composite widgets, and dynamic UI.

UI/UX

• UI Visual Quality Contract — Use before accepting visual UI changes.

Design System

• Design System Contract — Use when UI must follow reusable components, tokens, and interaction patterns.

Responsive UI

• Responsive Behaviour Contract — Use when layout must remain usable across viewports and content lengths.

Code Quality

• Code Quality Contract — Use for maintainability and structural acceptance.

Architecture

• Architecture Boundary Contract — Use when module, layer, dependency, or ownership boundaries matter.

Refactor

• Refactor Safety Contract — Use for behaviour-preserving structural change.

Clean-up

• Dead Code Removal Contract — Use when removing apparently unused files, exports, branches, or docs.

Documentation

• Documentation Truth Contract — Use when docs, examples, and guides must match real implementation.
• README Claim Contract — Use for README updates, public pages, and repository positioning.

Release

• Release Evidence Contract Plus — Use before merge, publication, release note approval, or public handoff.

Security

• Security Boundary Contract — Use when permissions, identities, integrations, data, or sensitive operations change.
• Secrets Handling Contract — Use for environment variables, credentials, CI, logs, tokens, and configuration.

Privacy

• Privacy and Data Contract — Use for personal, tenant, analytics, export, retention, or audit data.

Observability

• Logging Safety Contract — Use when adding or changing logs, metrics, traces, or alerts.

Reliability

• Error Handling Contract — Use when adding, changing, or reviewing failures, retries, fallbacks, and messages.

Performance

• Performance Budget Contract — Use when claims or changes affect performance, rendering cost, API use, or bundle weight.

Dependency

• Dependency Addition Contract — Use before adding packages, frameworks, tools, or external services.

API

• API Stability Contract — Use when public API, schema, route, exported function, or integration behaviour may change.

Delivery

• Git Hygiene Contract — Use when packaging branch changes, commits, generated files, or merge handoffs.

Template Product

• Template Library Quality Contract — Use for this repository’s prompt, skill, and contract assets.

Output

• Agent Output Format Contract — Use when agent handoffs must be structured, inspectable, and easy to review.